The Agent Skills Directory

Indirect Prompt Injection: The skill is designed to ingest and process user-supplied KQL queries and natural language descriptions to generate complex graph queries. Ingestion points: Untrusted user input is accepted in SKILL.md to define the structure and source data for graph construction. Boundary markers: The skill does not define specific delimiters or instructions to ignore potential commands embedded within the user-provided data. Capability inventory: The skill utilizes tools like kusto_query for executing commands (including database management) and provides instructions for generating local files and launching applications. Sanitization: There are no documented procedures for sanitizing or validating the input data before it is interpolated into generated queries.
Dynamic Execution and Command Execution: The skill provides a mechanism to generate and run local PowerShell scripts to support Windows-based visualization. Script Generation: The references/KUSTO_EXPLORER_LAUNCH.md file contains instructions for the agent to assemble a .kql file and execute PowerShell commands to locate and launch the Kusto Explorer desktop application. Safety Context: The skill includes explicit security guidance, such as avoiding PowerShell here-strings to prevent injection and requiring the use of ask_user for explicit consent before any file operations or process launches occur.

azure-kusto-graph